Basic processes in reading: Computation of abstract letter identities.

A speeded classification experiment examined the hypothesis that an early processing stage in reading involves the computation of abstract letter identities. When 20 undergraduates were asked to base their classification on physical criteria, letter strings that differed in case but shared the same letter identities (e.g., HILE/hile) were classified as "different" less efficiently than strings with a common phonological code but different spelling (e.g., HILE/hyle). Letter strings with a common phonological code but different spelling were classified as efficiently as letter strings without a common phonological code (e.g., HILE/hule). Results of the present experiment, along with other experimental findings and some neuropsychological observations, provide converging evidence for a representation and comparison process that is neither visual nor phonological but is based on abstract letter identities. It is suggested that the computation of abstract letter identities is a precursor to lexical access during reading. Implications for the interpretation of certain developmental reading difficulties are noted. (French abstract) (26 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)